Epsilon Archive for Student Projects

Abstract

This study in the subject of forest planning is made for Sveaskog and its aim is to develop tools that prognosticate outcome in terms of volumes per sort from not yet cut forest stands. The degree of detail is just to get reliable results for at least one months volumes, for Sveaskogs own forestland in the market region of Norrbotten that is about 100 000 m3fub. The course of action to accomplish this is regression analysis and the data consists of cut tracts on own forestland in Norrbotten during the years 2006, 2007 and 2008. That is a little more than 4 million m3fub.

Two tools were made. One that uses planned tracts data (PT) for computation and one that uses the stand register data (BR). The PT-tool has the best precision of basically two obvious reasons. First PT is further in the supply chain and therefore closer to the answer, the industry measuring. Second, BR had some pretty serious practical issues attached to it at this time.

An internal validation was made with all the data and it showed that the calculated total volume for PT was 0,2 % below the true volume and the same volume for BR was 4,7 % above. More detailed, roughly the PT-tool estimates all sorts but burning wood and GROT fair while the BR-tool estimates all sorts but pine and spruce timber poor.

Finally, the aim of the study is reached because the tools are completed and the PT-tool shows promising results for practical use while the BR-tool right now can be tested but later be redone with better structured data.